European societies are faced with growing inequalities (e.g., Cingano, 2014; Grabka & Frick, 2008; OECD 2015). While there seems to be a shared understanding that extreme levels of inequality lead to negative societal outcomes (Neckerman & Torche, 2007), this does not necessarily imply that adhering to strict equality is a strong consensus. Indeed, individuals’ views on the distribution of goods and burdens are more complex. Among others, the legitimacy of inequality is judged by the extent to which it corresponds to normative expectations individuals hold regarding the principles that should guide a just distribution. Justice theory has identified four such distributive principles, i.e., equality, need, equity[1], and entitlement. The evaluation of existing inequalities as legitimate by citizens is crucial. Indeed, it posits the link between the level of inequality and its speculated societal- and individual-level consequences as only inequalities that are evaluated as illegitimate are suggested to erode social cohesion and trust (e.g., Liebig & Sauer, 2016). Thus, it is essential to have relevant measurement tools to investigate which justice principles are used when judging distributions as just or unjust. Addressing this need, Hülle et al. (2018) developed the Basic Social Justice Orientations Scale (BSJO) to capture individuals’ agreement with each of the four principles and implemented the scale in the German research context. They showed that preference for equity, equality, need, and entitlement varies between individuals. Substantial variation, however, is also expected between societies as normative distributive preferences are speculated to vary with the basic structure of a society (Arts & Gelissen, 2001; Miller, 2001).

Distributive justice is concerned with the allocation of reward and burdens. Within distributive justice, two types of attitudes may be distinguished. Outcome-related justice evaluations refer to individuals’ evaluation of the result of a distribution process, i.e., an evaluation of what is. Order-related justice, on the other hand, refers to ideas about what a just distribution of rewards and burdens should look like within a social aggregate, i.e., what ought to be (Liebig & Sauer, 2016). The latter is the focus of this paper. More precisely, the focus is on the norms and principles according to which goods (e.g., wealth) and burdens (e.g., taxes) should be distributed within a society.

While closely related to ideas of inequality and redistribution, the order-related justice principles transcend the dichotomy of equality/inequality. Scholars of justice came to understand justice as a multi-faceted concept and they identified four distinct distributive principles that individuals refer to when evaluating distributions. Equity theory (Adams 1963) stated that justice was assessed through proportionality, or more precisely, on the grounds of proportionality of individual contributions and rewards received in return relative to other’s contributions and rewards. Other justice principles were later identified—equality, need, and entitlement (Deutsch, 1975; Hülle et al., 2018; Miller, 1979, 2001)—and allowed to examine the plurality of attitudes toward distributive justice. These four principles can be defined and distinguished as follows: (i) Equality stands for an equal distribution of rewards and burdens irrespective of individual characteristics and contributions. (ii) Equity suggests a distribution based on merit through the basis of proportionality between individual inputs and involvement on one side and the profits to be enjoyed on the other. (iii) The principle of need calls for a distribution that takes individual basic needs into account, while (iv) the entitlement principle[2] suggests that ascribed (e.g., gender) and achieved status characteristics (e.g., occupational prestige) should determine the distribution of goods and burdens (Deutsch, 1975; Hülle et al., 2018; Miller, 2001). It ought to be underlined that these four principles are not mutually exclusive (Van Hootegem et al., 2020). One can, for instance, be in favor of equity and need as enshrined in the “Boulding principle” (Boulding, 1962; Traub et al., 2005).

While individuals within a society differ in their agreement with the four principles (e.g., Hülle et al., 2018), what is regarded as just can also vary with the structure of a society (Miller, 1979). As argued by Wegener and Liebig (1995), different systems can produce different attitudes towards distributive justice. In addition to structural society-level characteristics, one’s own position in this society and prior experience of justice and injustice, can also shape perceptions of justice. This is what Wegener and Liebig (1995) distinguished as normative and rational argumentation. Normative argumentation indicates that “different societies have different dominant justice norms” (Wegener & Liebig, 1995) and rational argumentation states that the rational interest of individuals with respect to their own social position may influence their views on distributive justice[3]. The aim of the BSJO scale is to identify preference for the four distributive principles within and between countries.

Item generation and selection

The Basic Social Justice Orientation Scale (BSJO) (Hülle et al., 2018) was developed to measure people’s agreement with the four principles: equality, equity, need, entitlement. The original 12-item scale (as well as the 8-Item short scale) was developed and validated in the German context and has been fielded in a number of studies representatives of the German population (e.g., SOEP-IS, ALLBUS, LINOS). For the ESS Round 9 rotating module “Justice and Fairness in Europe,” a 4-item version of the BSJO scale was developed, including small adjustments to original question wording. Accordingly, each item serves as a manifest indicator of one justice principle. Table 1 reports the final question wording for all four items provided in the ESS source questionnaire which serves as the basis for translation into all languages used in the ESS data collection[1].The ESS R9 offers the first cross-country data source to study agreement with the four distributive principles equality, equity, need, and entitlement. To facilitate substantive research on the four principles, we address concerns of measurement quality in this report. In doing so, we acknowledge that using single-item indicators to represent complex latent concepts—while often necessary in large-scale survey programs—suffers from methodological shortcomings (Rammstedt & Beierlein, 2014; Van Hootegem et al., 2021). With respect to the basic social justice orientations, shortcomings introduced by the one item per dimension operationalization implemented in the ESS R9, may be especially problematic given recent evidence from Germany and Belgium. Van Hootegem et al. (2021) highlight discrepancies between the single-item measures (as used in ESS R9) and the latent operationalization of the four principles as proposed by Hülle et al. (2018). In relying on the four single-item measures to capture respondents’ attitudes about basic social justice orientations, particular attention should thus be given to concrete item formulations and their overlap with the underlying theoretical concept. Differences between single-item measures included in the ESS R9 and the underlying theoretical concept of order-related justice, for example, include that the latter deals with the allocation of goods and burdens more generally, while the four items of the BSJO are only concerned with the allocation of goods and privileges. Even more specifically, the items measuring equality and equity limit their focus on the concept of income (or income and wealth in case of the equality item). This focus on income and wealth may affect how positively or negatively respondents’ evaluate the four principles. This is particularly relevant considering that preference for distributive principles may differ depending on which resource (or burden) is being distributed and between whom (Konow, 2003; Van Hootegem et al., 2020). Accordingly, the specific wording of each of the four BSJO items should be taken into consideration, both in interpreting our findings as well in using the four BSJO items in subsequent research.

The ESS places a strong focus on cross-cultural questionnaire development, i.e., during the development process, experts from all ESS languages and cultural contexts are regularly consulted, and both quantitative and qualitative pre-testing is conducted in multiple language-country contexts. The questionnaire development process - including selection of and changes to the BSJO items - is documented in the module design documents available from www.europeansocialsurvey.org. Moreover, translation procedures followed the TRAPD protocol developed for cross-cultural survey research (Harkeness, 2003).

Samples

For our main analyses, we rely on data from ESS Round 9 fielded in 2018/2019 covering 29 European countries (ESS Round 9: European Social Survey Round 9 Data 2018). The ESS is a cross-sectional face-to-face survey conducted every 2 years in Europe. Country samples are representative of all individuals aged 15 and over that are posed of a fixed core program and rotating modules. R9 features questions on distributive and procedural justice that are part of the newly developed rotating questionnaire module “Justice and Fairness in Europe.” The four items of the BSJO scale, intended to measure preference for the four distributive principles equality, equity, need, and entitlement, were fielded among all respondents.

Accordingly, our analyses rely on 49,519 observations with country samples ranging between 861 respondents in Iceland and 2,745 respondents in Italy. We complement our analyses with data from the Cross National Online Survey (CRONOS), which featured select questions of the “Justice and Fairness” module later fielded in ESS R9 (CROss-National Online Survey panel 2018).

CRONOS is an online survey conducted among respondents of ESS R8 in Great Britain, Slovenia, and Estonia. CRONOS is a panel study re-interviewing respondents in six different waves of the survey from 12/2016 to 02/2018 (Villar et al., 2018). For the purpose of this report, we use information from Wave 3 and Wave 6. Country-level analyses rely on information on country-level unemployment in 2018 sourced from Eurostat, available in the multi-level data set provided by the ESS (Version 1.0).

Analyses are conducted in Stata 17 and employ the analytical weights (anweight) provided in the ESS data (Kaminska 2020).

Item analyses and item parameters

Response distribution for the four distributive principles across the entire sample of 29 European countries are displayed in Figure 1 in the original publication[2]. On average, agreement is highest for the equity item with an overall mean of 3.96 (ranging between 3.79 in Finland and 4.27 in Austria) and the need item with an overall mean of 3.89 (ranging between 3.35 in the Czech Republic and 4.10 in Austria). On average, respondents in the 29 European countries disagree with the entitlement item with an overall mean of 2.16 (ranging between 1.84 in Iceland and 2.89 in Slovakia). Agreement is mixed for the equality item, with country averages ranging between 2.62 in Norway and 3.93 in Italy and an overall mean of 3.39. This greater between-country variation on the equality item is reflected in the highest intraclass correlation (ICC) of 0.11 compared to low between-country variance on the equity item (0.02) and moderate between-country variance on need (0.05), and entitlement (0.08). In line with this observation, within-country variance is generally lowest for the highly supported equity item and the need item as indicated by low standard deviations and left-skewed distributions (see Figures A1-A4 and Table A1 in the Appendix in the original publication). Generally, these summary statistics suggest that there seems to be a consensus within and between European societies that equity and need are guiding principles for a just society, while there is less agreement with regard to the principles of equality and entitlement.

Item non‑response

A high share of non-response can indicate substantial problems on the side of respondents in understanding and/or responding to a survey question and may introduce bias in subsequent analyses. Therefore, we investigate the share of missing values due to item non-response across all four items in all 29 countries. For the total sample, non-response rates are low and range between 1.7% for the equity item and 3.0% for the entitlement item. A closer look at non-response rates on the country level (see Table A1 in the Appendix in the original publication) reveals that in the great majority of countries non-response levels are below 5% for all items. Entitlement is the item with the highest share of non-response in most countries. Bulgaria shows the highest shares of missing values on all four items, with non-response reaching up to 17.8% for the entitlement item, casting severe concern about the measurement quality of the distributive justice principles in Bulgaria.

Objectivity

The 4-item version of the BSJO ensures objectivity of application through a standardized questionnaire format and labeled response categories.

Reliability

The four BSJO items were fielded twice in the CRONOS panel among the same set of respondents. Respondents from Estonia, Great Britain, and Slovenia first answered these questions in CRONOS Wave 3 (fielded June– August 2017) and then again as part of Wave 6 (fielded January–February 2018), allowing to assess reliability of the four BSJO items[1]. Table 2 in the original publication reports average agreement with the four order-related justice principles in ESS R9 and the CRONOS panel. Comparing the average agreement across CRONOS and ESS R9 for Estonia, Great Britain, and Slovenia, we find that, on the country level, agreement is largely consistent across time points and studies. In particular, the order of preference for each principle remains stable across studies and measurement time points. The panel data structure of the CRONOS study further allows investigating within-person stability. Estimating intraclass correlations (ICC)[2] (Koo & Li, 2016) as a measure of test-retest reliability, we observe ICCs ranging from 0.32 to 0.65. Compared to reliability standards recommended for measures used in diagnostics (Koo and Li, 2016), these values indicate low test-retest reliability. However, keeping in mind that, first, the  time-span  between  repeated  measures  was  quite  long,  and that, second, not much is known about the stability  of distributive preferences, and, third, that low ICC values may be the result of low variance in underlying values and a low number of measurement time points (Koo & Li, 2016), we may not compare  the  observed  test- retest reliability to quality thresholds commonly used in  the diagnostics literature but rather compare ICC across countries and items (also see Berchtold, 2016). In doing so, we find the test-retest reliability for the equality item to be slightly lower in Slovenia compared to Great Britain and Estonia. Across countries, test-retest reliability is highest for the need and entitlement principles and lowest for the equity item. 

The share of cases in percent, where responses in Wave 3 and Wave 6 of CRONOS matched perfectly, also reported in Table 2 in the original publication, further underlines this conclusion. Among the four BSJO items, the share of perfectly matched responses is highest for the equity item in Estonia and Britain, and second-highest in Slovenia. After recoding information from the 5-point response scale into three categories that only differentiate between “Agreement,” “Neither/nor,” and “Disagreement,” the share of consistent responses ranges between 60 and 82%. This underscores that the four BSJO items allow to reliably identify individuals who agree or disagree with each principle but may be less reliable in assessing the degree of (dis)agreement.

Future research should address the question of stability of distributive preferences—in perhaps long-spanning panel data—more thoroughly. Recent inclusion of items measuring attitudes towards the four distributive principles in the Socio-Economic Panel study in Germany will facilitate this type of research (Adriaans et al., 2021).

Validity

We report associations with related variables spanning socio-demographic characteristics, attitudes on redistribution and income inequality, and the country context.

Variables:

Socio demographics Following the notion that preferences for distributive principles reflect positional effects, i.e., that they are shaped by one’s own position in society, we study the association of age, gender, years spent in full-time education, household income, employment status, and perceiving oneself to belong to a discriminated group with the preference for order-related justice principles. Household income refers to household net income and allowed respondents to place themselves in 1 of 10 income bands reflecting country-specific income deciles. Information on employment status corresponds to the main activity the interviewee has been doing for the last 7 days and indicates if a respondent is in paid work or not. Respondents further stated if they belonged to a discriminated group.

Attitudes towards inequality and political attitudes. Moreover, the following attitudinal measures were used to assess if normative preferences indeed map onto corresponding attitudes towards income inequality, redistribution, and political preferences, underlining concurrent validity of the four justice principles. Respondents evaluated the fairness of very high and very low incomes within their country. The question provided information on actual income levels in decile 10 and decile 1 in a country’s income distribution and used a 9-point response scale ranging from − 4 “extremely unfair, unfairly low incomes” to 4 “extremely unfairly, unfairly high incomes,” with 0 labeled as “fair incomes”5. Respondents also reported their agreement with the following statement serving as a measure of preference for redistribution: “The government should take measures to reduce differences in income levels”, where 1 refers to “Disagree strongly” and 5 refers to “Agree strongly.” Interviewees further reported their political orientation on a left-right scale, where 0 was labeled “left” and 10 was labeled “right.”

We provide evidence on construct validity of the four BSJO items by means of studying correlations with related concepts. As findings on the social distribution of normative justice principles are scarce, we link our findings to the existing literature on the preference for redistribution. Generally speaking, equality and need may be thought of mechanisms reducing inequality and equity and entitlement as principles that legitimize inequality (Hülle et al., 2018).

Socio‑demographics. Table 3 in the original publication shows correlations between socio-demographic characteristics and the four items in the pooled sample. Correlations by country are presented in the Appendix (Table A2) in the original publication. Figure 2 further plots average agreement with the four items by gender, membership in a discriminated group, and employment status[3]. In the pooled sample, we observe small positive correlations of equality and need with age, i.e., older respondents tend to show higher agreement with these principles and lower agreement with the entitlement item. Contrary to these patterns Hülle et al. (2018) report that in Germany agreement with the need principle is highest among young respondents and Olivera (2015) reports that, after controlling for cohort effects, younger individuals are more in favor of redistribution. Forsé and Parodi (2009), however, report only small differences by age group with older respondents reporting stronger agreement with equality and equity.

Following (Hülle et al., 2018) and (Forsé and Parodi, 2009) who report a negative relationship between education and equality, we find that respondents who spent more years in full-time education tend to agree less strongly with the equality item showing a moderate correlation (r = − 0.19).

Moreover, agreement with the equality item is negatively related to household income, underscoring the notion that individuals’ preferences for distributive principles are not independent from self-interest (Hülle et al., 2018; Reeskens & Van Oorschot, 2013). In line with findings from Rehm (2009) who found that unemployment levels at the occupational level increase preference for redistribution, respondents who are not in paid work, are more in favor of equality (see Fig. 2 in the original publication). Somewhat surprisingly, considering that equity is the dominant distributive norm in the context of work (e.g., Adams, 1963; Deutsch, 1975), agreement with the equity item does not differ considerably between respondents who are in paid work and those who are not. Again, following the notion that one’s own position and experiences may shape distributive preferences, respondents who state they are a member of a discriminated group report stronger agreement with the need item and lower agreement with the entitlement item.

Empirical evidence further suggests that women are more in favor of justice principles reducing inequality, rather than principles that legitimize it (Forsé & Parodi, 2009; Hülle et al., 2018; Liebig & Krause, 2006; Wegener & Liebig, 2000). Small yet consistent mean differences are found in ESS R9. Women show stronger agreement with the equality item and lower agreement with the equity item.

Overall, observed correlations and mean differences are small to moderate suggesting that agreement with the four BSJO items does not reflect mere positional effects but rather complex normative preferences that transcend self-interest and one’s current position within a countries’ social structure.

Attitudes towards inequality and political stance. The four distributive justice principles refer to normative ideals about how goods and burdens should be allocated within a society. Accordingly, we would expect agreement with these principles to predict evaluations of the existing income distribution, where very low and very high incomes violate both equality and need principles. On the other hand, the principles of equity and entitlement may legitimize the observed inequalities suggesting that respondents may evaluate the income distribution as more fair. ESS R9 respondents evaluated both top (Decile 10) and bottom incomes (Decile 1) within their country. In line with expectations, respondents who show stronger agreement with the equality and need items evaluate bottom incomes as more unfairly low and top incomes as more unfairly high (Table 4 in the original publication). Also, in line with expectations, respondents with stronger agreement with the entitlement item evaluate top incomes as less strong overreward and bottom incomes as less severe underreward. The association between equity and evaluations of top and bottom incomes shows the reversed association. While at first glance this may seem counter- intuitive, equity only legitimizes inequality that is due to differences in effort. Accordingly, very low incomes of full-time workers and very high incomes that may be due to factors other than “hard work” violate the equity norm. Upon closer inspection, in Denmark, Finland, Iceland, Norway, and Sweden—most of which are characterized by low levels of actual income inequality—the correlation between equity and evaluation of top incomes is positive and the correlation between equity and evaluation of bot- tom incomes is negative, suggesting that in these settings very low and very high incomes may be viewed as violating the equity norm. Contrary to this observation, agreement with the equity norm shows the expected negative correlation with the evaluation of top incomes in 17 out of the 29 countries studied (Table A2 in the Appendix in the original publication).

We grouped the four principles into mechanisms that may reduce inequality (equality, need) and those that may legitimize inequality (entitlement, equity), suggesting a close link between distributive justice and preference for redistribution. If individuals are in favor of the equality principle, we would expect them to also support government intervention in reducing inequalities. The other principles are less clearly linked to preference for redistribution by means of government intervention but the categorizing of principles into inequality-reducing and inequality-legitimizing suggest a positive association with need and negative correlations with equity and entitlement. The strong and moderate positive correlations observed for preference for redistribution with the equality and need items as well as the moderate negative correlation with the entitlement item are in line with this expectation. In the pooled sample, agreement with the equity item shows a weak yet positive correlation with preference for redistribution. Closer inspection of correlations by country shows a heterogeneous picture with both positive and negative effects that seem to follow a North-South divide with negative effects in Denmark, Finland, Great Britain, Iceland, Netherlands, Norway, and Sweden, and positive effects in Bulgaria, Cyprus, Croatia, Czech Republic, France, Hungary, Italy, Latvia, Lithuania, Montenegro, Poland, Portugal, Serbia, Slovakia, and Slovenia. Explanations for these heterogeneous patterns may relate to country differences in the extent to which individuals perceive redistribution through government intervention as a valid measure to address existing inequity in the distribution of income (Ahrens, 2019). Interestingly, there seems to be an overlap between the countries where we observed positive correlations between agreement with the equity item and the evaluation of more pronounced injustice at the top and bottom of the income distribution. At the same time, these countries (Denmark, Finland, Iceland, Sweden, and Norway) are characterized by high tax burdens and extensive welfare states (Kautto, 2010) which suggests that, among those with higher support of the equity norm in these countries, (additional) government intervention is notices in the income distribution.

Correlations between equality, need, entitlement, and the preference for redistribution vary in size but not in direction across countries. Beyond the political means of redistribution, the political spectrum corresponds to ideas of a fair society. In line with the pro-redistribution stance from the political left (Bean & Papadakis, 1998) and that economic equality is related to a left-leaning political preference in Europe (Hadarics, 2017), respondents tending towards the left in their political self-evaluation show stronger agreement with equality and need, and less agreement with equity and entitlement as guiding principles of a just society. These patterns are fairly consistent across countries.

Overall, correlations between the four principles and related attitudes such as the preference for redistribution, political orientation, and the justice evaluation of the income distribution are in line with expectations providing supportive evidence that the four BSJO items measure the underlying constructs of equality, equity, need, and entitlement.

Structural conditions. Besides identifying individual agreement with the distributive principles of equality, need, equity, and entitlement, the four items of the BSJO implemented in ESS R9 are also intended to identify between country differences in the preferences for norms that guide the allocation of goods and burdens within a society. As stated by Wegener and Liebig (1995), structural conditions at the country level are expected to translate into different justice norms. Olivera (2015), for example, showed that the unemployment rate is associated with increased demands for redistribution. A similar pattern can be observed for the principles of equality and need (see Fig. 3 in the original publication). Overall, agreement with equality and need items is higher in countries with higher unemployment rates. That is, in countries were objective need is higher, the inequality reducing principles need and equality find higher agreement.

Beyond correlations with broader conditions on the labor market, it has been argued that ideas of distributive justice are enshrined in different types of welfare states (Clasen & van Oorschot, 2002; Sachweh, 2016). Building on an extension by Eikemo et al. (2008) of the classic welfare state typology suggested by Esping-Andersen (1990), Fig. 4 in the original publication shows profiles of agreement with the four order-related justice principles by welfare state type (group means and 95% confidence intervals are also shown in Table A4 of the Appendix in the original publication)[4]. In line with the response pattern that was already introduced above, the equity item finds strong agreement in all welfare state types, with the lowest agreement (3.81) across the Anglo- Saxon countries (i.e., Great Britain and Ireland) but over- all low between-country variation. In line with Clasen and van Oorschot (2002), agreement with the equity item is especially pronounced in Bismarckian welfare states. The entitlement item, on the other hand, meets disagreement across the board with the lowest level of support (1.9) in the Southern welfare states. Consistent with the particular strong disagreement with the inequality-legitimizing entitlement norm as a guiding principle for a fair society, Southern European countries show the strongest agreement with the inequality-reducing equality and need principle. Support for the need principle is also very pronounced in the Scandinavian and Bismarckian welfare states but finds less agreement in the Eastern European countries. The equalityitem, finds weak support in Bismarckian, Anglo-Saxon, and Eastern European countries and moderate support in Southern European countries. Only in Scandinavia, we observe overall disagreement with the equality item which may seem surprising given that the Scandinavian welfare state is characterized by universalism (Clasen and van Oorschot 2002). In combination with the positive correlation observed for unemployment and equality, this may suggest that agreement with the order-related justice principles does not necessarily reflect which distributive norms are enshrined in the countries’ institutions but rather reflect which principles would address current inequalities. When analyzing country-level patterns, it should be noted that the four distributive principles do not directly map onto specific welfare policies as welfare states may differ in their interpretation of distributive principles into policies (Clasen & van Oorschot, 2002). For example, the definition of “basic needs,” whose coverage is demanded by the need principle, may vary substantially between more or less generous welfare states, even if the principle of need receives similar levels of popular support. This may also help to explain why observed differences between welfare state types are rather small (while statistically significant). Moreover, the analyses is restricted to the European con- text. Follow-up research that extents its sample beyond European countries may shed more light on how structural conditions across more heterogeneous institutional settings relate to preferences for distributive principles. 

Descriptive statistics

See Figures A1-A4 and Table A1 in the Appendix in the original publication.

Jule Adriaans, Bielefeld University, Faculty of Sociology, Universitätsstraße 25, 33615 Bielefeld, Germany, e-mail: jule.adriaans@uni-bielefeld.de

The data used for this study stem from the European Social Survey and the CROss-National Online Survey. ESS data is available free of charge through the ESS Data Portal (https://ess-search.nsd.no) and CRONOS data is available through the ESS website. Code for all analyses presented in the paper are available here https://osf.io/ankum/.